Hello,
We have five hardware PRTG servers(one master, 4 probes and 4000 sensors overall) and I'm shuffling servers with theirs sensors around in order to find the best performance for our PRTG setup.
1. What parameter to monitor and compare on the probes(if CPU and MEM look the same)? I'm comparing the quantity of established TCP sessions. What is YOUR general recommendations? When the load is heavy, WMI sensors go crazy, as you know.. 2. I have a probe with 400 sensors(WMI, pings, HTTP) and another one with 1000 sensors(pretty the same set). And still, the second one has twice less connections established than the first one... Intervals are the same. Why that can be? Yes, I've rebooted the servers already :)
Thanks a lot. Mike
Hello,
unfortunately for WMI there can't be a general rule of thumb, as it heavily depends how fast the targets respond (and also how fast they can verify monitoring requests with their Domain Controllers). This can mean a Probe with only a handful of targets, that respond very slow, could cause much more troubles, than a Probe with hundreds of fast responding targets.
Technically this is an issue on all protocols, but it's very dominant on WMI based sensors.
best regards.
Mar, 2014 - Permalink